Solution-Processable Singlet Fission Photovoltaic Devices - Nano

Dec 17, 2014 - This is the first solution-processable singlet fission system that performs with ..... Photovoltaic Devices Based on Colloidal PbX Quan...
0 downloads 0 Views 314KB Size
Letter pubs.acs.org/NanoLett

Solution-Processable Singlet Fission Photovoltaic Devices Le Yang,† Maxim Tabachnyk,† Sam L. Bayliss,† Marcus L. Böhm,† Katharina Broch,† Neil C. Greenham,† Richard H. Friend,† and Bruno Ehrler*,†,‡ †

Cavendish Laboratory, J J Thomson Avenue, Cambridge CB3 0HE, United Kingdom Center for Nanophotonics, FOM Institute AMOLF, Science Park 104, 1098 XG Amsterdam, The Netherlands



S Supporting Information *

ABSTRACT: We demonstrate the successful incorporation of a solution-processable singlet fission material, 6,13-bis(triisopropylsilylethynyl)pentacene (TIPS-pentacene), into photovoltaic devices. TIPS-pentacene rapidly converts highenergy singlet excitons into pairs of triplet excitons via singlet fission, potentially doubling the photocurrent from high-energy photons. Low-energy photons are captured by small-bandgap electron-accepting lead chalcogenide nanocrystals. This is the first solution-processable singlet fission system that performs with substantial efficiency with maximum power conversion efficiencies exceeding 4.8%, and external quantum efficiencies of up to 60% in the TIPS-pentacene absorption range. With PbSe nanocrystal of suitable bandgap, its internal quantum efficiency reaches 170 ± 30%. KEYWORDS: Singlet fission, TIPS-pentacene, colloidal nanocrystals, hybrid photovoltaics, multiple carrier generation, quantum dots inglet fission is an exciton multiplication mechanism in organic molecules, converting one singlet exciton into a pair of triplet excitons. It can occur on an ultrafast time scale (pico- or even femto-second) when the triplet state has approximately half the singlet energy, out-competing singlet decay via fluorescence.1,2 Singlet fission has recently attracted considerable attention because it provides a pathway to overcome the single-junction Shockley-Queisser limit in solar cells by reducing thermalization losses.2,3 To benefit from singlet fission, a larger-bandgap singlet fission material has to be coupled to a smaller-bandgap semiconductor that acts as an electron acceptor and generates a single exciton per absorbed low-energy photon.4,5 Pentacene has been the first singlet fission material to be successfully implemented in working solar cells, with external quantum efficiencies (EQEs) of up to 126% in pentacene/C60 solar cells when light trapping was employed, with internal quantum efficiencies (IQEs) approaching 200%.5−7 Bilayer cells with pentacene and infrared-absorbing inorganic nanocrystals as electron acceptors have been demonstrated with PbS and PbSe. The power conversion efficiency (PCE) of the PbS cells reached nearly 1%;8 and with PbSe, the PCE improved to 4.7%.9,10 The EQE obtained in these devices peaked at 80%, though the EQE at pentacene absorption peak (670 nm) was 30%.9 Despite these encouraging results, the instability in air and light hinders the commercial prospects of solar cells made from pentacene. Pentacene undergoes photo-oxidation, forming an endoperoxide as oxygen bridges across the C6 and C13 positions that are most susceptible to oxidation.11−13

S

© XXXX American Chemical Society

Pentacene is insoluble in most common solvents so it needs to be deposited by thermal evaporation under vacuum. The 6,13-bis(triisopropylsilylethynyl)pentacene (TIPS-pentacene) was originally designed for good charge mobility in field-effect transistors,14−17 and it shows a dramatically enhanced solubility and oxidative stability in solution.18,19 Besides protecting the susceptible C6 and C13 positions with the bulky TIPS groups,20 it has been suggested by Fudickar and Linker that they also help to ensure cycloreversion of the endoperoxide to the parent acene, preventing its further decomposition.13 Maliakal et al. demonstrated theoretically that the TIPS groups lower the triplet energy of the acene, preventing singlet oxygen formation, and that they also lower the LUMO energy, minimizing charge transfer.11,12 However, empirically both its triplet energy and LUMO were not found to be lower than unsubstituted pentacene.21 In concentrated solutions, TIPS-pentacene shows a triplet yield of 200%, with singlet fission occurring via a transient excimer intermediate, formed when one photoexcited molecule collides with a ground-state molecule.21 In thin films of TIPSpentacene, singlet fission is found to occur on a time scale of 1 ps with a triplet yield of 144%.22 However, solar cells based on TIPS-pentacene have so far delivered limited promise. Lloyd et al. achieved a PCE of 0.52% with solution-processed TIPSpentacene/C60 bilayer devices.23 A separate study on vacuum Received: September 23, 2014 Revised: December 6, 2014

A

dx.doi.org/10.1021/nl503650a | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

components do not react with each other, which is in contrast to TIPS-pentacene and fullerenes.32,33 An inherent shortcoming of this device structure is that light is incident from the bottom of the substrate, passing through the nanocrystal layer first where up to 85% of the light is absorbed. This diminishes the EQE contribution from TIPSpentacene. Using a transparent top and reflective bottom contact can potentially circumvent this limitation (see detailed discussion in the Supporting Information Section 3), as TIPSpentacene receives a larger fraction of light as it is incident from the top. Surface ligands on the nanocrystals are crucial for charge transport and interfacial properties with acenes.34,35 The long, insulating native ligands (oleic acid) on the nanocrystals are exchanged for shorter, more conductive ligands in a solid-state process.36 Common ligands such as tetrabutylammonium iodide (TBAI), 1,2-ethanedithiol (EDT), and benzene-1,3dithiol (BDT) have been used as shorter ligands for improved charge transport.30 Even though the heterojunction between TIPS-pentacene and all the aforementioned combinations led to devices with considerable performance and charge generation from both materials (Supporting Information Section 4), subsequent experiments and optimization were done with BDT-capped PbSe and PbS for consistency and ease of fabrication. Figure 2 shows the power conversion efficiency (PCE) and open-circuit voltage (Voc) of nanocrystal/TIPS-pentacene devices for a range of nanocrystal sizes (PbSe in Figure 2a and PbS in Figure 2b). All parameters are enhanced with larger bandgap for each series. The increase in open-circuit voltage (Voc) with increasing nanocrystal bandgap in both series corresponds to the increase in diagonal bandgap between the donor−acceptor layers. Together with the higher fill factor measured (see Tables S2 and S3 in the Supporting Information), this raises the PCE of the device. Under various optimized conditions, Voc as high as 0.60 V and short-circuit current (Jsc) up to 21.4 mA/cm2 have been achieved with a champion cell producing a PCE of 4.8% (Figure 2c, and see Section 5 of Supporting Information for full solar cell characterization). Figure 2d shows a typical EQE spectrum of a PbSe-only device in comparison with a PbSe/TIPS-pentacene device. The underlying shape of the PbSe/TIPS-pentacene spectrum follows that of the PbSe-only spectrum, as the majority of light is being absorbed and converted into photocurrent by the PbSe (70−85% in the range 500−700 nm). The peaks seen at 591 and 648 nm follow the features of the absorption of a TIPS-pentacene film, indicating that there is significant photocurrent contribution from TIPS-pentacene over the spectral region from ∼550−700 nm. It is noteworthy that the EQE over this region reaches up to 60%. Interestingly, we only see photocurrent generation from the TIPS-pentacene features at 648 and 591 nm, but not at 692 nm. This peak at 692 nm has been assigned to an aggregate feature of TIPS-pentacene film due to improved packing of the molecules.21,37 In photoluminescence (PL) measurements on thick neat TIPSpentacene films, its emission peak occurs at 667 nm (Supporting Information Figure S3). The absence of an emission from the lowest-energy 692 nm peak, whose absorption is associated with aggregates,21,37 indicates that these states efficiently recombine nonradiatively, explaining their absence in the EQE spectra of all the devices.

sublimed TIPS-pentacene/C60 devices showed a PCE of 0.42%.24 Meanwhile, colloidal nanocrystals have garnered widespread interest in solar cells of various designs,25−29 recently reaching a PCE of 8.55%.30 By careful control of reaction temperature and time, nanocrystal size can be easily tuned, which inversely correlates to the bandgap energy. Monodispersity is crucial for good performance to give an optimal compact arrangement of the nanocrystals and a clean energy landscape.31 Small-bandgap nanocrystals absorb well into the red and near-infrared regions, capturing low energy photons and making them ideal electron acceptors for low-energy TIPS-pentacene triplet excitons. Here, we report the incorporation of solution-processed TIPS-pentacene into photovoltaic devices with small-bandgap PbS or PbSe nanocrystals as the acceptor material, obtaining a PCE of 4.8% and an EQE of up to 60% in the region where TIPS-pentacene absorbs. We find that the IQE exceeds 100% for the TIPS-pentacene layer, unless the energetics of the nanocrystals do not allow for triplet exciton dissociation, in which case the IQE nearly halves. The nanocrystal/TIPS-pentacene bilayer device architecture is shown in Figure 1. Electrons are transported through TiO2

Figure 1. (a) Device architecture of the nanocrystal/TIPS-pentacene photovoltaic device. (b) Chemical structure of TIPS-pentacene. (c) Alignment of energy levels in the PbS/TIPS-pentacene device. Blue arrows illustrate the singlet fission process in TIPS-pentacene upon photoexcitation. When PbSe is used instead of PbS, this energy diagram remains similar except for the valence band of PbSe being at −5.1 eV, and its conduction band varies accordingly. TIPS-pentacene HOMO and nanocrystal valence bands were determined by ultraviolet photoelectron spectroscopy (UPS) as described by Ehrler et al.9

and extracted from ITO, while holes are extracted through the gold top contact (Figure 1c). Contrary to previous pentacene/ nanocrystal bilayer architectures,8−10 TIPS-pentacene needs to be deposited on top of the cross-linked nanocrystals layer due to a lack of orthogonal solvents. We find that the thicknesses of individual neat films add up to that of the bilayer film. Furthermore, the absorption peak of the PbS nanocrystals remains unchanged when TIPS-pentacene is spun on top (Supporting Information Figure S4). This suggests that the two B

dx.doi.org/10.1021/nl503650a | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 2. Device PCE and Voc variation with bandgap of (a) PbSe and (b) PbS in the nanocrystal/TIPS-pentacene devices, all fabricated under similar conditions. Unfilled data points represent the best performance under these conditions. (c) The I−V characteristics of the 1.25 eV PbS/TIPSpentacene champion cell (upon further optimization, specifically of the duration of BDT soaking time, see Supporting Information Section 1.3 for details) with PCE exceeding 4.8% in comparison to a device made from TIPS-pentacene and PbSe and a device made from PbSe only. (d) EQE spectrum of a PbSe/TIPS-pentacene device compared to that of a PbSe-only device of the same bandgap.

Figure 3. (a) Energy alignment between the TIPS-pentacene triplet state and the charge transfer (CT) states with different nanocrystals bandgaps (in brackets), where the CT energy is approximated by the difference between the nanocrystal conduction band energy and the TIPS-pentacene highest occupied molecular orbital (HOMO). (b) EQE contribution and modeled absorbed light fraction from TIPS-pentacene. (c) IQE of TIPSpentacene in bilayer devices with nanocrystals of different bandgaps, isolated from total EQE by transfer matrix optical modeling.

The presence of a positive EQE contribution from TIPSpentacene in all the devices suggests that the conduction band edges of all the PbSe/PbS used in our study are low enough to ionize the triplets generated from singlet fission in TIPSpentacene, leading to net electron transfer from TIPSpentacene to the nanocrystals. Analogous to previous studies on PbSe/pentacene bilayer devices,9 we expect a negative contribution toward the overall EQE from TIPS-pentacene when the bandgap of the nanocrystal is sufficiently large such that the conduction band level is too high to ionize the triplets (Figure 3a). Thus, when TIPS-pentacene is combined with 1.3 eV bandgap PbSe nanocrystals, this EQE contribution is reduced, with the remaining contribution presumably from the low-bandgap tail of the nanocrystal distribution and possibly from some direct singlet exciton transfer prior to singlet fission (Figure 3b).

To investigate the origin of the TIPS-pentacene photocurrent, transfer matrix optical modeling was performed to calculate the absorption profile of the device which then allows the extraction of the internal quantum efficiency (IQE) (Supporting Information Section 6).7,38 It is assumed that the heterojunction consists of a sharp and flat interface between the layers. We first modeled the PbSe-only IQE from the measured EQE of PbSe-only devices. This IQE can then be used to model the PbSe EQE component of the overall EQE in a PbSe/TIPSpentacene device. The difference between the measured overall EQE of a PbSe/TIPS-pentacene device and this PbSe component represents the TIPS-pentacene component of the EQE (Figure 3b), because PbSe and TIPS-pentacene are the only active layers contributing toward photocurrent. The IQE of TIPS-pentacene can be obtained by using its isolated EQE contribution divided by its absorbed light fraction (eq 1). C

dx.doi.org/10.1021/nl503650a | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

photocurrent change is not indicative of the fraction of triplets contributing to the photocurrent, as the majority of the photocurrent is produced by the nanocrystals, which are slightly different in the two devices. For larger bandgap nanocrystals, we observe no photocurrent change within the resolution of our experimental setup (